Severe developmental letter-processing impairment: A treatment case study

This article was downloaded by: [McGill University Library]On: 18 December 2014, At: 08:51Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office:Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Cognitive NeuropsychologyPublication details, including instructions for authors and subscriptioninformation:http://www.tandfonline.com/loi/pcgn20

Severe developmental letter-processingimpairment: A treatment case studyRuth Brunsdon a b , Max Coltheart a & Lyndsey Nickels aa Macquarie University , Sydney, Australiab Children's Hospital at Westmead , Sydney, AustraliaPublished online: 23 Apr 2007.

To cite this article: Ruth Brunsdon , Max Coltheart & Lyndsey Nickels (2006) Severe developmentalletter-processing impairment: A treatment case study, Cognitive Neuropsychology, 23:6, 795-821, DOI:10.1080/02643290500310863

To link to this article: http://dx.doi.org/10.1080/02643290500310863

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”)contained in the publications on our platform. However, Taylor & Francis, our agents, and ourlicensors make no representations or warranties whatsoever as to the accuracy, completeness, orsuitability for any purpose of the Content. Any opinions and views expressed in this publicationare the opinions and views of the authors, and are not the views of or endorsed by Taylor &Francis. The accuracy of the Content should not be relied upon and should be independentlyverified with primary sources of information. Taylor and Francis shall not be liable for anylosses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilitieswhatsoever or howsoever caused arising directly or indirectly in connection with, in relation to orarising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any substantialor systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, ordistribution in any form to anyone is expressly forbidden. Terms & Conditions of access and usecan be found at http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/loi/pcgn20

http://www.tandfonline.com/action/showCitFormats?doi=10.1080/02643290500310863

http://dx.doi.org/10.1080/02643290500310863

http://www.tandfonline.com/page/terms-and-conditions

Severe developmental letter-processing impairment:A treatment case study

Ruth BrunsdonMacquarie University, Sydney, Australia and Children’s Hospital at Westmead, Sydney, Australia

Max Coltheart and Lyndsey NickelsMacquarie University, Sydney, Australia

We report the case of ET, a 7-year-old child with a severe developmental letter-processingimpairment. Detailed assessment revealed multiple impairments of the letter-processing systemaffecting abstract visual letter representation, semantics, and connections between the phonologicalprocessing modules and the orthographic buffer. Treatment methods focused on letter soundingthrough development of abstract visual representation and semantic representation of letters.Treatment resulted in dramatic and enduring improvements in cross-case matching, letter/numbercategorization, and letter sounding.

Cognitive neuropsychological reading research has largely focused on word processing, with a rela-tive neglect of single-letter processing. The current study investigated letter processing in greaterdetail than has been usual and outlines a broad theoretical framework for letter processing. ET’sassessment and treatment data are used to support and question predictions from the framework.A number of theoretical implications are discussed with reference to ET’s data and that of otherreported cases of impaired single-letter processing. Finally, the paucity of investigation of letter pro-cessing in children is highlighted, particularly with regard to integrity of abstract letter representationin developmental dyslexia.

INTRODUCTION

Adults with normal language functioning are ableto process words in all modalities. This is reflectedin cognitive neuropsychological models of langu-age that represent word (and nonword) processingin terms of comprehension and production ofwritten and spoken words (e.g., Ellis & Young,1996). Single letters can also be processed in

different modalities. For example, literate adultscan recognize and name written letters, writesingle letters to dictation, and comprehendspoken letter names and sounds. They can alsocomprehend words when their letter names arespelled aloud, and they can spell words orally bynaming component letters serially. However,theoretical consideration of single-letter proces-sing across modalities is rare.

Correspondence should be addressed to Ruth Brunsdon, Rehabilitation Department, Children’s Hospital at Westmead, Locked

bag 4001, Westmead, NSW, 2145, Sydney, Australia (Email: [email protected]).

Ruth Brunsdon was supported by a Department of Education, Science and Training Australian Post Graduate Award and

Lyndsey Nickels by an Australian Research Council QEII Fellowship during the preparation of this paper.

We are indebted to our client ET and his family for their enthusiastic participation throughout the project.

# 2006 Psychology Press Ltd 795http://www.psypress.com/cogneuropsychology DOI:10.1080/02643290500310863

COGNITIVE NEUROPSYCHOLOGY, 2006, 23 (6), 795–821

Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

Letter-processing research has largely focusedon single-letter identification in reading. Thus,most experimental investigations and theories ofsingle-letter processing are centred around visualletter perception or recognition (e.g., Golden,1986; Massaro, 1998; Massaro & Hary, 1986;McClelland & Rumelhart, 1981; Morrison &Butler, 1986; Mozer, 1989; Polk & Farah, 1997;Sanocki, 1988, 1991a, 1991b) and in the contextof a letter string (e.g., Besner, Coltheart, &Davalaar, 1984; d’Ydewalle & Auwers, 1994;McClelland & Rumelhart, 1981; Jacewicz, 1979;Sanocki, 1988). There are very few studies thatinvestigate visual letter processing in isolation(Arguin & Bub, 1995; Jacobs & Grainger, 1991).In addition, research rarely includes detailed inves-tigation of single-letter processing in nonvisualmodalities, such as auditory perception of letternames and sounds or semantic representation ofletter knowledge.

Although most individuals with acquiredlanguage impairment exhibit core deficits inword processing (and not letter processing), anumber of case studies have now been reportedwith core impairments at the level of letter proces-sing: for example, impairments affecting earlyvisual perception of letters (Lambon Ralph &Ellis, 1997), abstract letter identification(e.g., Behrmann & Shallice, 1995; Greenwald &Gonzalez Rothi, 1998; Howard, 1987; Kay &Hanley, 1991; Miozzo & Caramazza, 1998), orwritten letter production (e.g., Del GrossoDestreri, 2000; Hanley & Peters, 1996). In par-ticular, recent interest in the role of letter recog-nition in pure alexia has resulted in a number ofstudies investigating the integrity of abstractletter representation in this disorder (e.g., Arguin& Bub, 1994; Behrmann & Shallice, 1995; Kay& Hanley, 1991; Reuter-Lorenz & Brunn,1990). Cases have also been reported with differ-ent acquired impairments for letters from thosefor words. These include an inability to namesingle letters but preserved ability to spell wordsorally (Greenwald & Gonzalez Rothi, 1998), anintact ability to read many words aloud in thecontext of severe impairments in matching lettersof different cases (Howard, 1987), or severe

impairments in letter naming and sounding butrelatively intact reading of words aloud (Sevush& Heilman, 1984). In summary, even in thewell-researched realm of adult language processingour knowledge of how single letters are processedremains limited.

We know even less about the role of letterprocessing in developmental dyslexia. Very littleis known about how letter-processing skills areacquired in normal development or how impair-ments in letter processing impact on readingdevelopment. It is this that forms the focus ofthe current study. We first report a single casestudy of the assessment and treatment of severedevelopmental letter-processing impairments in ayoung child. We then set out a proposed theoreti-cal framework that offers an account of letterprocessing and consider our case study in thelight of this framework.

CASE REPORT

At the time of assessment and treatment ET was7 years old. He attended a mainstream primaryschool. He was referred to the first author by histreating neuropsychologist for a detailed assess-ment of reading skills. Both ET’s mother andschool staff had become increasingly concernedabout his inability to associate written numbersand letters with their names or sounds, andparticularly his failure to make improvements inreading despite intensive input.

ET had a long history of language delay, beha-vioural immaturity, and attention difficulties. Atage 3 a formal diagnosis of language delay wasmade by a speech pathologist, and ET thenattended speech therapy until school age. Hearingwas assessed as normal. There was a strong familyhistory of language-based learning difficulties(those affected included his mother and sister).

Formal intellectual assessment was first con-ducted at age 4 and revealed low average intellec-tual development, across both verbal and nonverbaltasks. Attention difficulties and emotional labilitywere also noted. Significant difficulties withattention and concentration were consistently

796 COGNITIVE NEUROPSYCHOLOGY, 2006, 23 (6)

BRUNSDON, COLTHEART, NICKELS

Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

reported by preschool and primary school staff,and ET was commenced on stimulant medicationat age 5.

Neuropsychological assessment was conducted 6months prior to the current investigation.Intellectual skills fell in the low average rangeoverall (see Table 1, Differential Ability Scales,GCA). Spatial skills were average. Nonverbalreasoning and verbal skills fell in the lowaverage range. Further assessment of language andattention revealed specific difficulties in both areas(refer to Table 1: WISCIII, digit span; NEPSY,speeded naming and comprehension of instructions;TEA-Ch, DT (divided attention).

At the time of the assessment ET’s markeddifficulties with reading, spelling, and numberskills were the main concern and particularly hiscontinuing inability to learn alphanumericsymbols. He was reported to have severe problemswith letter and number processing in comparisonto class peers. Intensive intervention had beenimplemented at school throughout kindergartenin the areas of rhyming, phonological segmentation,and letter writing and sounding. In Year 1 specificintervention focused on specific letter-to-soundtraining using picture cues, reading using onsetrhyme techniques, and sight word training.Improvements were noted in phonological skillsincluding segmentation and blending, but accord-ing to his special education teacher letter namingand sounding remained very poor and inconsistentdespite constant teaching and practice.

Pretreatment assessment

Pretreatment assessment was conducted over 3months. ET was aged between 7:0 and 7:3 atthe time of assessment and was in his finalterm of Year 1 (2nd year of attendance) at school.The assessment focused on letter processing. Alsoincluded was a brief assessment of word recognitionand semantics, picture naming, general phonologi-cal processing, and number processing. Controldata were collected for selected letter-processingtasks. There were two unimpaired control subjects(C1 and C2): C1 was age 7:11 and in Year 2, andC2 was age 6:9 and in Year 1.

Table 1. Neuropsychological assessment results

Task

Standard Score/T-score/

Scaled score

Differential ability scales

(Elliot, 1990)

Standard

scoresa

General conceptual ability (GCA) 88

Verbal 85

Nonverbal reasoning 83

Spatial 101

T scoresb

Recall of designsb 49

Pattern constructions 53

Similarities 39

Word definitions 44

Matrices 41

Sequential and quantitative reasoning 39

Matching letter-like forms 66

Wechsler intelligence scale for children (3rd Edition) (WISCIII)

(Wechsler, 1991)

Scaled

scorec

Digit span

Forward span raw score ¼ 5.

Backward span raw score ¼ 0.

5

NEPSY

(Korkman, Kirk, & Kemp, 1997)

Scaled

scorec

Phonological processing 8

Speeded naming 2

Comprehension of instructions 4

Verbal fluency 9

TEA-Ch

(Manly, Robertson, Anderson, & Nimmo-Smith, 1999)

Scaled

scorec

Sky search (measure of selective

and focused attention)

Number 8

Time 7

Attention score 7

Score! (measure of sustained attention) 7

Sky search DT (measure of

sustained-divided attention)

1

Note: Results are for ET aged 612 years.

aStandard score: mean of 100 and standard deviation of 15.bT score: mean of 50 and a standard deviation of 10.cScaled score: mean of 10 and standard deviation of 3.

COGNITIVE NEUROPSYCHOLOGY, 2006, 23 (6) 797

TREATMENT OF LETTER-PROCESSING IMPAIRMENT

Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

Language and semanticsGeneral semantic knowledge and language skillswere screened using spoken word–picture match-ing, spoken picture naming (PALPA subtests 47and 53; Kay, Lesser, & Coltheart, 1992) and theTest of Reception of Grammar (TROG; Bishop,1983). ET’s performance on spoken word–picture matching was almost perfect (38/40), andhe achieved a standard score of 80 (low average)on the TROG. However, ET only scored 28/40on spoken picture naming (errors are shown inAppendix A), a poor performance for his age.1

Most of ET’s errors were semantic errors. ET’smother also reported that he made occasionalsemantic errors in conversation.

Phonological processing: Repetition, segmentation,and blending skillsGeneral phonological skills were assessed usingphonological segmentation of initial and finalsounds (PALPA subtests 16 and 17), theChildren’s Test of Nonword Repetition (CNRep;Gathercole & Baddeley, 1996) and PhonologicalProcessing (a phonological segmentation task)from the NEPSY (Korkman, Kirk, & Kemp,1997). ET’s performance indicated generallylow average (but not impaired) phonologicalprocessing. ET made one error on the PALPAsegmentation of initial-sounds tasks (22 itemsadministered) and three errors on the segmenta-tion of final-sounds tasks (14 items administered).ET achieved an age scaled score of 7 on NEPSYphonological processing and a standard score of89 on nonword repetition (CNRep).

Phoneme blending was assessed using a pro-nounceable nonword blending task (auditory pres-entation). Stimuli consisted of a three-phonemenonword list (adapted from the Coltheart andLeahy (1996) word/nonword test) and a two-phoneme nonword list devised by the authors.2

The examiner sounded aloud each phoneme(at one phoneme per second; e.g.,/æ//t//i:/),

and ET was required to blend them togetherto make the corresponding nonword (“attee”).ET’s phoneme blending was excellent for two-phoneme nonwords (10/10). His blending ofthree-phoneme nonwords (4/10) was poor, butlimited auditory working memory skills (seeTable 1, digit span backwards) may have been acontributing factor.

Word recognitionVisual lexical decision. A visual lexical decision taskwas devised incorporating 15 very-high-frequencywords (e.g., on, day, make), 15 legal nonwords(e.g., im, gop, lish), and 15 illegal nonwords(e.g., td, kfe, hwci). ET was required to circlethe words. His response accuracy was 73%(33/45 correct). He identified 7/15 of the wordsand made four false positive errors includingthree legal nonwords and one illegal nonword.

Spoken- to written-word matching task. In this taskthe target stimuli were 10 very-high-frequencywords. ET had to match the spoken stimulus toone of four response choices, including the target(e.g., take), a visually similar word (e.g., make), avisually dissimilar word (e.g., when), and a visuallysimilar nonword (e.g., toke). ET chose the targetword correctly on 4 of the 10 trials, which wasno better than chance (binomial, p ¼ .15). Errorsincluded two nonword responses and four visuallysimilar word responses.

Letter naming and soundingET’s performance was not significantly affected bythe use of standard fonts (such as that used inPALPA, Kay et al., 1992) over the New SouthWales Foundation Script (font taught in primaryschools in the Australian State of New SouthWales). Thus for most assessment tasks standardPALPA stimuli or Times New Roman font wereemployed (for author-developed tests).

1 Two 6-year-old controls scored 36/40 and 37/40 on this task (control data collected for a previous study: Brunsdon, Nickels,

Coltheart, & Joy, 2004).2 The three-phoneme list consisted of equal numbers of consonant–vowel–consonant (CVC) and VCV combinations. The

two-phoneme list consisted of equal numbers of CV and VC combinations.



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

Letter naming: PALPA Test 22. ET was able toprovide the names of 6/26 lower case letters (a fi m o s) and 7/26 upper case letters (A E I O SV X). He made a total of 39 errors including 37nonresponses (i.e., “don’t know”). Two errorresponses were number names (d! “four”,l! “one”).

Letter sounding: PALPA Test 22. ET was able toprovide the appropriate phoneme for five lowercase letters (f i m o s), but no upper case letters.Again most errors were nonresponses (37/47).Other error responses consisted of target letternames (three for lower case and six for uppercase stimuli) and one incorrect letter sound(d! /b@/).

Spoken to written matching of letters: PALPA Test23. Matching spoken letter names and sounds to

written letters was also poor but well abovechance (forced-choice task with four responsechoices). When provided with spoken letternames ET matched 19/26 correctly (binomial,p ¼ .000) and with spoken letter sounds matched13/26 correctly (p ¼ .004).

Letter writingWriting letters to dictation. ET was able to write5/26 lower case letters (l i a m o) and 3/26 uppercase letters (E, I, S) to dictation of letter names.Errors were mostly nonresponses (41/52). On theupper case task other errors included the lowercase version of the correct response and one incor-rect lower case letter. For the lower case task, errorresponses included incorrect lower case letters andone number. ET produced 6/26 correct writtenletters to dictation from sounds (a d m o s x). Allerrors were nonresponses.

Figure 1. Samples of free writing.



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

Free writing. Two samples of free writing areshown in Figure 1. For both samples ET wassimply asked to write some words and letters(ET wrote his name three times, which has notbeen shown). For the second sample, ET wasasked to read back his writing.

In summary, language comprehension and phono-logical processing (including nonword repetition)fell in the low average range, consistent with lowaverage intellectual development. But significantdifficulties were evident in picture naming, witha predominance of semantic errors.

Although ET was able to recognize some very-high-frequency words in a lexical decision task andspoken to written matching task, his performanceremained very limited for his age. Thus, at thesingle-word level, ET’s development of ortho-graphic representations was severely delayed.

Finally, ET also displayed a severe impairment inletter naming and sounding and in writing letters todictation. However, further assessment was necess-ary to determine his exact pattern of intact andimpaired processing at the level of single letters.

Cognitive neuropsychological assessment ofletter processing:

1. Phonological processing of letter namesand sounds

Letter repetitionET was able to repeat letter names and soundswith no errors.

Auditory letter name decision taskStimuli consisted of 26 letter names (e.g., /ef/,/di:/) and 26 nonword distractors (1–2 phonemes,e.g., /ep/, /fu:/). ET was asked “which of thesehave you heard before, which are the names ofletters?” His performance for letter names wasperfect and there was only one false positive error.

Spoken production of letter namesET was able to spontaneously produce all letternames, as demonstrated by his intact ability torecite the alphabet.

In summary, phonological processing of letternames/sounds (including auditory perceptionand spoken production) appeared relatively intact.

2. Visual letter perception and recognition

Matching identical written stimuliMatching of identical (same case and font) writtenletters was perfect (see Table 2).

Perception of letter orientationPerception of letter orientation was assessed usingthree tasks.

Mirror Reversal Task 1: PALPA Test 18. Taskadministration was altered slightly due to ET’stendency to be overwhelmed by simultaneouspresentation of multiple stimuli. Each item wasshown one at a time, and ET was required to indi-cate which stimuli were in the correct orientation(i.e., “the right way round”). He made four errors(see Table 2), which although below the levelof controls, represented no more than a mildimpairment.

Mirror Reversal Task 2. Because PALPA Test 18only uses 12 letters of the alphabet, a second taskwas devised that included an equal number oflower and upper case letters and included moreof the letters of the alphabet. Letters that areperfectly symmetrical were excluded (e.g., O,

Table 2. Visual letter perception and recognition: Proportion correct

Task ET C1 C2 N

Matching identical letter stimuli

Upper case 1.0 — — 15

Lower case 1.0 — — 15

Mirror Reversal Task 1

(PALPA Test 18)

.88 1.0 .97 36

Mirror Reversal Task 2 .86 — — 28

Letter orientation task .96 — — 26

Cross-font matching .95 — — 22

Letter/pseudoletter decision .89 1.0 .97 75

Cross-case matching

(PALPA Test 19)

.62 1.0 1.0 26

Note: ET: Age 7:0–7:3; Year 1, Term 4. C1: Age 7:11; Year 2,

Term 2. C2: Age 6:9; Year 1, Term 2.



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

I, T). ET again performed relatively well withequal success for upper and lower case stimuli,making only four errors in total (see Table 2).

Letter orientation task. This task was also devisedby the authors. ET was given a letter and wasasked to put it in the correct orientation. Theletters were presented to him either in thecorrect orientation or rotated by 90, 180, or 145degrees. ET’s performance was almost perfect(see Table 2).

Thus, ET’s early visual analysis of letters appearedrelatively intact. He was able to match identicalletter stimuli with ease and to represent letterorientation relatively well.

Abstract visual letter representation independent offont, but not caseCross-font matching. Cross-font matching wasassessed using contrasting fonts. ET was requiredto match the target letter to two versions of thesame letter in different fonts (but the same case),which were presented among six distractors. Hehad no difficulty with this task (see Table 2).

Visual letter/pseudoletter decision task. This taskrequired ET to distinguish between real Englishletters (N ¼ 25) and other letter-like symbols(e.g., Arabic, Hebrew, and Greek script, N ¼ 50)presented one at a time and all matched asclosely as possible for complexity. ET was requiredto indicate which stimuli he recognized. ET made

89% correct choices (he identified 24/25 of the realEnglish letters and made nine false positive errors).Although this performance was lower than that ofcontrols, his ability to recognize English lettersappeared relatively intact (see Table 2).

Abstract visual letter representation independent offont and caseCross-case matching was assessed using PALPATest 19 and was no better than chance (binomial,p ¼ .079). Controls performed perfectly on thistask. ET’s cross-case matching responses werenot based on the visual similarity between upperand lower cases of a letter. Even for those letterswith identical upper and lower case forms (e.g.,O and o or S and s) he performed at chance (46%).

Thus, ET appeared to be able to abstract acrossdifferent fonts but was not able to abstract lettersacross case.

3. Written letter production

Letter copyingUpper and lower case letters. Letter copying wasperfect for both upper and lower case letters (seeTable 3).

Delayed copying. ET only made one error on thedelayed copying task (see Table 3). Each letterwas displayed for 5 seconds then ET was askedto draw a circle and then write the target letterin the circle.

Table 3. Written-letter production: Proportion correct

ET C1 C2

Task Proportion Timea Proportion Timea Proportion Timea N

Letter copying upper and lower case 1.0 — — 34

Delayed copying .95 — — 22

Letter completion .76 — — 50

Copying English vs. Greek letters

English 1.0 36 1.0 25 1.0 33 13

Greek .77 68 1.0 46 1.0 61 13

Cross-case copying 0.0 — — 20

Note: ET: Age 7:0–7:3; Year 1, Term 4. C1: Age 7:11; Year 2, Term 2. C2: Age 6:9; Year 1, Term 2.aIn seconds.



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

Thus, peripheral writing mechanisms requiredfor direct and delayed copying were intact.

Letter completion. ET was presented with incom-plete letters (both upper and lower case) andasked to complete them. He was able to completea large proportion of the letters, and given the dif-ficulty of the task this was considered a relativelygood performance (see Table 3).

English versus Greek letter copying. Stimuli consistedof 13 upper case English letters and 13 Greek lettersmatched as closely as possible for complexity. ETwas asked to copy both sets of letters in turn asfast as possible. His copying was immaculate withonly three minor errors made on the Greek script.However, ET was twice as slow copying the unfa-miliar Greek script as when copying the Englishletters (see Table 3). The difference in copyingspeed between English and Greek stimuli was com-parable to that of the control subjects.

Cross-font copying. It was noted during same-casecopying that ET copied two of the target lettersin a very deliberate way (e.g., Q ! ), so acopying task was designed that required ET tocopy a number of different font versions of thesame letter. ET on nearly all occasions produceda generic version of the target letter rather thancopying each letter exactly.

These three copying tasks (letter completion,English vs. Greek letter copying, cross-font copying)require access to a relatively abstract written letterform. ET’s normal performance on these tasks (inconjunction with his production of a number ofcorrectly formed letters in free writing) suggestsrelatively good production of written letters.

Cross-case copying. Cross-case copying was impos-sible for ET (see Table 3). Despite promptingand teaching ET repeatedly provided same-casecopies of the stimuli. ET was also required to

cross-case copy letters that he had successfully pro-duced in free writing. For example, ET producedan A, n, and T in free writing. He was then pre-sented with the opposite case (e.g., a, N, & T)in a cross-case copying task, to determinewhether he could produce the target letters inresponse to their opposite case. He performedvery poorly with only 3/12 correct responses. Insummary, cross-case copying was impossible forET, in the context of otherwise relatively goodwritten letter production (as discussed above).

4. Letter semantics

Knowledge about lettersET was asked a number of questions about letters.Questions were considered to be easy for his age.3

Questions were forced choice (e.g., “Which lettercomes first as you say the alphabet, B or X?” or “Isthe letter S a vowel or consonant?”). He answered6/13 questions correctly, which was no better thanchance (p ¼ .209). The control subjects answered11/13 (C1) and 13/13 (C2) correctly. Whenasked to provide a definition of a letter ET replied“used for people . . . to help people”. When furtherprompted he replied “helping people . . . helpingpeople write”. The first control subject’s definitionwas “something you use to write with” and thesecond control said “an/eI/ or /1ks/ or /waI/ orsomething like that” and then with promptingadded “we use them to spell our name and writewords”. In sum, ET’s general knowledge aboutletters seemed extremely poor for his age.

Categorizing letters and numbersET was required to indicate whether an alphanu-meric stimulus (12 numbers and 12 letters) was anumber or a letter. On all three tasks (writtenlower case, 58% correct; written upper case, 58%correct; and auditory presentation of letternames, 42% correct) his performance was nobetter than chance (written upper and lower case,p ¼ 0.117; auditory, p ¼ .117).

3 Due to differences in exact curriculum between schools the Special Education teacher at ET’s school was also contacted and was

provided with a list of the questions used. She indicated that all of the questions would be answered correctly by ET’s peers, with the

exception of the distinction between vowels and consonants, which would be answered correctly by most.



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

Letter/number confusion errors were alsoevident in a spoken to written matching task.This task consisted of 40 stimuli (22 letters and18 numbers). ET had to match the spoken stimu-lus to one of four response choices (included 2numbers and 2 letters). Although his performancefell above the level of chance (63% correct,p ¼ .000) his errors were consistent with aninability to distinguish between numbers andletters. That is, a total of 12 errors were betweencategory confusions. In summary, ET could notdistinguish between numbers and letters, whenpresented in the visual or auditory modality.

Letter-processing assessment summary

ET displayed an overall severe impairment in letternaming and sounding and in writing letters todictation. Specific assessment of letter perception,letter semantics, and letter production revealed thefollowing relatively intact skills:

1. Visual perception of letter forms: as shown byintact matching and copying of identicalletters (same case and font) and intactperception of letter orientation.

2. Abstract visual representation of letters inde-pendent of font: as shown by intact matchingof same letters in different fonts, intact cross-font copying, and ET’s relatively good perform-ance on a visual letter/pseudoletter decision task.

3. Phonological processing of single-letter namesand sounds: as shown by intact repetition ofletter names and sounds, intact recitation ofthe alphabet from memory, and ET’s abilityto discriminate easily between recognizableletter names and other similar-lengthnonwords (that are not letter names).

4. Written letter production: as shown by ET’sability to complete partial letter forms, his pro-duction of a number of letters in free writing(plus more in other writing to dictationtasks), and his faster copying of Englishletters than Greek letters.

In contrast the following skills were impaired:

1. Abstract visual representation of letters inde-pendent of case: as shown by ET’s inability tomatch or copy letters across case.

2. General knowledge (or semantic represen-tation) of letters: as shown by ET’s inabilityto categorize letters as vowels or consonantsand his impoverished general knowledgeabout letters. ET’s poor letter semantics couldrepresent a category-specific semantic impair-ment given his normal performance onPALPA spoken word to picture matching.However, given that this test can be insensitiveto semantic impairment (Cole-Virtue &Nickels, 2004) and ET’s frequent semanticerrors in naming and conversation, we suspectthat his letter semantic difficulties in fact formpart of a more general semantic impairment.4

3. Writing letters to dictation of names andsounds: presumably due to a general failure tolearn associations (or form connections)between intact spoken and written letter forms.

Number processing

ET had significant number-processing impair-ments, and their pattern mirrored his letter-processing impairments. In brief, number copyingand matching were perfect as was repetition ofnumber names. However, naming writtennumbers was virtually impossible (even for thenumbers 1–10). ET had very impoverishedknowledge about numbers and was unable to com-plete simple mental number sums. He was unableto write numbers to dictation.

TREATMENT STUDY

Treatment focus

Treatment was conducted during the first twoterms of Year 2 (ET’s 3rd year of school atten-dance). The overall aim of treatment was toenable ET to associate written single letters with

4 Thus, we consider letter semantics to be part of the general semantic system. We propose that letters form one of the many

dissociable semantic subcategories.



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

their associated phonemes (phonemes were tar-geted as letter sounds are more functional thanletter names for learning to read). Treatmentmethods reflected ET’s core level of impairment.Thus, both upper and lower case letters weretaught in letter pairs, and extensive semanticelaboration (through the use of stories, pictures,and songs) was incorporated into treatment foreach target letter. The treatment aims were to:

1. Improve ET’s ability to sound letters inresponse to a visual letter stimulus

2. Improve ET’s ability to associate both upperand lower case versions of each letter with asingle sound.

3. Improve ET’s ability to categorize letters andnumbers.

Treatment design

Treatment outcome measures included lettersounding (lower and upper case), letter/numbercategorization (auditory and written), and cross-case matching. For these outcome measures, twopretreatment assessments were conducted (a thirdpretreatment assessment was conducted for lettersounding). Pretreatment assessments 1 and 2 wereconducted 2 months apart. The third pretreatmentassessment was conducted 1 month after the second.

The 26 target stimuli consisted of all of the lettersof the alphabet. The 26 target letters were dividedinto two treatment sets (Sets A and B), each of 13letters. The two sets were matched as closely aspossible on letter frequency5 (Mann–Whitney,p ¼ .762) and on the visual similarity betweenupper and lower case6 (Mann–Whitney, p ¼ .960;Boles & Clifford, 1989). The two sets were alsomatched according to letter-sounding accuracy oneach pretreatment assessment. Treatment wascommenced following the final pretreatment assess-ment. As shown in Table 4, Set A was targeted for

treatment first. Midtreatment assessment was con-ducted after completion of Set A treatment (Set Atreatment was discontinued after the midtreatmentassessment) and prior to treatment of Set B. Alloutcome measures were assessed at midtreatment,immediately following treatment of Set B and 8weeks later (follow-up assessment). A range ofother skills were also assessed (relating to letterand word recognition) following treatment and arediscussed below.

Treatment methods

Each set of 13 letters was treated identically. Eachset was treated over 5 weeks. During Weeks 1, 2,and 3 three new letters were introduced. Foreach week, on Days 1 and 2, the first letter waspractised, on Days 3 and 4 another letter was prac-tised, on Days 5 and 6 the third letter was prac-tised, and finally on Day 7 all three letters werereviewed. During Week 4, four new letters wereintroduced, and Week 5 consisted of reviewingall 13 letters. ET’s mother was provided withdetailed instructions and conducted the home-based treatment daily.

The treatment programme aims required afocus on cross-case matching and semanticrepresentation. Treatment therefore employed a

Table 4. Treatment design

Time of assessment Type of assessment

Late October Pretreatment Assessment 1

Late December Pretreatment Assessment 2

Late Jan Pretreatment Assessment 3

February/March (5 weeks) TREATMENT OF SET A

Early March Midtreatment assessment

March/April (5 weeks) TREATMENT OF SET B

Early May Posttreatment assessment

Late June Follow-up assessment

5 Letter frequency was calculated by examining all the words in the DRC model’s monosyllabic vocabulary (Coltheart, Rastle,

Perry, Langdon, & Ziegler, 2001). The raw frequency of occurrence was determined for each letter, and this raw frequency was

divided by the total number of letters in the database.6 Boles and Clifford (1989) devise a number of visual similarity ratings for letter comparisons (both between different letters and

within letter cross-case comparisons). The ratings were devised by asking participants to rate similarity on a scale LOW ¼ 1 to

HIGH ¼ 5. The data used for the current study were the average ratings for cross-case comparisons (incorporating both Aa and

aA presentation order).



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

method that incorporated explicit pairing of lowerand upper case versions of each letter as well asextensive semantic elaboration to help create astrong and elaborate semantic representation ofeach letter. The treatment programme employedthe “Letterland” flashcard stimuli for lower andupper case letters (Manson & Wendon, 1997).In the Letterland system each letter has a charactername. For example, the letter h is known as “HarryHat Man”, and the capital letter H is presented asHarry Hat Man doing a hand stand (see Figure 2).Each letter has the Letterland character pictoriallyincorporated into the letter form and also an expla-nation of the capital letter shape (which is alwaysthe same letter character doing something differ-ent). Each flashcard has the letter with pictorialcharacter on one side and the letter presented inisolation on the other side. Other Letterlandmaterials employed in the current treatmentincluded the Letterland ABC Book (which con-tains a short story about each character) and theLetterland CD (which has a short song linkingeach letter character to its sound, set to acommon nursery rhyme tune). These materialswere incorporated into a daily treatment pro-gramme for ET. Each day ET and his mother(see Appendix B) targeted a particular letter.They read the letter character story, looked atthe lower case flash card, sang the song, generatedwords beginning with the letter sound, and thenlooked at the upper case flash card. They finishedeach daily practice session with a brief revision ofthe target letter and letter sound. ET thoroughlyenjoyed the programme and was keen to practiseeach day, reminding his mother if she forgot orwas busy doing other things.

Treatment study results:7

1. Outcome measures

Letter soundingThere was no significant change in letter sound-ing across the three pretreatment assessments

(overall: Cochrans Q, p ¼ .178; upper case,p ¼ .097; lower case, p ¼ .607; Set A, p ¼ .247;Set B, p ¼ .607). The only significant differencebetween upper and lower case letter soundingoccurred at the first pretreatment assessment(Mann–Whitney, p ¼ .020), thus lower case andupper case sounding data were combined for exam-ination of overall treatment success (see Figure 3).

Set A letter sounding improved dramaticallyafter treatment was targeted at Set A (posttreatmentof Set A when compared to pretreatment assessment2; McNemar, p ¼ .000). This improvement wasmaintained posttreatment of Set B (McNemar,p ¼ .375) and at follow-up (see Figure 3).

Set B letter sounding was not significantly betterfollowing treatment of Set A (McNemar, p ¼ .219,when compared to pretreatment assessment 2), butdid improve significantly after being targetedspecifically for treatment (McNemar, p ¼ .004;posttreatment of Set A vs. posttreatment of SetB). Treatment effects were maintained at follow-up.

Cross-case matchingThere was no significant change in cross-casematching across the two pretreatment assessments(overall: McNemar, p ¼ 1.0; see Figure 4). ET’sresponse accuracy was not significantly related toletter frequency or visually similarity betweenupper and lower case for either pretreatmentassessments (pretreatment assessment 1, visualsimilarity, Mann–Whitney, p ¼ .225, letter

Figure 2. Letterland “Harry Hat Man”. The original was in colour

and all occurrences of the letter “H” or “h” were printed in green.

7All comparisons between pre- and posttreatment results employ the highest pretreatment assessment result for establishing

improvements in total correct.



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

frequency, p ¼ .065; pretreatment assessment 2,visual similarity, p ¼ .052, letter frequency,p ¼ .562). Examination of his performanceacross pretreatment assessments 1 and 2 indicatedthat for 10/26 letters ET’s cross-case matchingwas consistently correct. Although, this indicatedwell developed cross-case representation for someletters, his performance was still well below ageexpectations. Note, again his pattern of res-ponding across pretreatment assessments (i.e.,consistently correct vs. consistently incorrect vs.inconsistent) was not significantly related toletter frequency (Kruskal Wallis, p ¼ .557) orvisually similarity (p ¼ .255).

Following treatment of Set A there was a sig-nificant improvement in cross-case matching(overall, McNemar, p ¼ .016, pretreatment 2 vs.posttreatment of Set A), reflecting equal improve-ment in Set A and Set B items (see Figure 4).In other words both Set A and Set B improvedalmost to ceiling level after treatment of onlyletters from Set A. Treatment effects weremaintained at follow-up.

Letter/number categorizationLetter/number categorization was assessed usingboth visual stimuli (upper and lower case) and audi-tory stimuli (letter names).8 Visual lower case and

Figure 3. Letter sounding: Proportion correct. Figure 4. Cross-case matching: Proportion correct.

8 In this task stimuli consisted of 12 letters and 12 numbers. Half of the letters were from Set A and half from Set B.



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

auditory stimuli were assessed twice prior to treat-ment, but only a single pretreatment assessmentwas conducted for visual upper case stimuli (seeFigure 5). There was no significant change in per-formance across pretreatment assessments (visuallower case, McNemar, p ¼ .754; auditory,McNemar, p ¼ .727). A significant improvementin performance was evident after treatment of SetA letters (visual lower case, p ¼ .002; visual uppercase; p ¼ .012; auditory, p ¼ .001). That is, per-formance approximated ceiling after treatment ofonly half of the target letters. Treatment effectswere maintained after treatment of Set B and atfollow-up (see Figure 5).9

2. Other measures

Posttreatment performance on other measures ofletter processing and word reading are shown inTable 5. Significant improvements were evidentacross a range of tasks not specifically targeted fortreatment (see Table 5 for p values). Matchingspoken letter sounds to written letters improvedsignificantly (this is not surprising, given that treat-ment aimed to associate written letters with theirletter sounds). There was also generalization oftreatment effects to written-letter production,and particularly writing letters to dictation (ofletter sounds) and cross-case copying. In addition,ET’s whole-word reading improved. ET wasunable to read any words other than his name con-sistently prior to treatment. After treatment he wasable to correctly read 14 high-frequency words(e.g., if, up, not, dog, box, dig, sun) and clearlydemonstrated an ability to grapheme sound andphoneme blend when reading. In addition, he pro-duced more words in free writing (see Table 5).

Some tasks did not change after treatment. Letternaming did not improve. When attempting letternaming ET offered the sound first, though withprompting was able to offer some names. Finally,encyclopaedic knowledge about letters (such as theorder of letters in the English alphabet and vowel/consonant status) did not improve. Of note,

however, was a better definition of a letter.Posttreatment ET provided the following definition(when asked “what is a letter?”): “Letters are used forlearning. I mean like learning how to do thething . . . like how to learn/s/”. Then whenprompted he continued “we make the ABC out ofletters . . . . we see letters on paper and in booksand we say the words . . . they are good to learn”.

3. Results summary

The results clearly indicate a specific treatment effectfor all outcome measures: letter sounding, cross-case matching, and letter/number categorization.

Figure 5. Letter/number categorization: Proportion correct (visual

upper case not assessed at pretreatment assessment 2).

9 At posttreatment a single assessment of categorization of auditorily presented letter sounds and number names was also

conducted (not shown in Figure 5 and not tested pretreatment). ET’s performance was perfect.



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

Treatment effects were maintained at follow-upassessment 8 weeks after cessation of treatment.

Results indicate some generalization of treat-ment effects to untreated stimuli—that is, theimprovement of Set B stimuli following treatmentof only Set A stimuli (before specific treatmentof Set B items). Treatment generalization wasevident for cross-case matching and letter/number categorization with performance on SetB items improving to ceiling levels followingtreatment of Set A. There was no treatmentgeneralization effect for letter sounding.

Post hoc analyses indicated smaller but signifi-cant improvements in high-frequency wordreading and written-letter production, includingcross-case copying. No significant change wasevident in naming written letters, or aspects ofspecific letter knowledge such as letter order inthe alphabet and vowel versus consonant status.

GENERAL DISCUSSION

While conducting the current study, and particu-larly when interpreting ET’s pattern of letter-processing impairment, it became clear thattheoretical models of language processing wereunable to provide an adequate framework. Suchmodels focus on word processing and fail to expli-citly incorporate processing of single lettersbeyond the level of visual perception. Much ofthe detail regarding where letters are representedand how they are processed is not illustrated ordescribed. In addition, current language-processing frameworks and/or research providevery little information about how aspects of letterprocessing develop and how they might beimpaired in developmental dyslexia.

Thus much of the current discussion is focusedon the proposal of a letter-processing framework,which is used to interpret ET’s assessment data

and treatment results. The framework is shownin Figure 6 and is based on the assumption thatprocessing of single letters and words occurs inthe same system. The core word-processing com-ponents are drawn from commonly used modelsof language processing (e.g., Coltheart, 1987;Ellis, 1982; Ellis & Young, 1996).10 The additionsand amendments specific to letter processing arisefrom the following question: Can existing modelsof language processing explain all aspects ofsingle-letter processing, and, if not, what additionalcomponents might be required? The frameworkexplores letter processing beyond the level of per-ception and thus also provides a conceptualizationof auditory letter processing, letter naming, letterwriting, and letter semantics.

Single-letter processing—A preliminaryframework:

1. Phonological processing of letter namesand sounds

Auditory recognition of letter names occurs viathe auditory analysis system and phonologicallexicon. The auditory analysis system (seeFigure 6) allows auditory perception of letternames and sounds. The phonological lexiconenables auditory discrimination between letternames and nonwords—that is, it representsnames of letters (and words) that we havelearned to recognize. Retrieval of letter names(e.g., saying the alphabet) occurs when thespoken form of the letter name is activated in thephonological lexicon, and spoken productionoccurs via the phoneme production system.11

Nonwords (and letter sounds) can be repeatedvia the direct link between the auditory analysissystem and phoneme production system.

If we examine ET’s performance in relationto the framework shown in Figure 6 we can con-clude that his auditory analysis, phonological

10 We have chosen to employ a version of this model with a single orthographic lexicon (e.g., Jackson & Coltheart, 2001).

However, we acknowledge that ET’s pattern of impairment can be explained adequately by this version or by a version with separate

orthographic input and orthographic output lexicons (e.g., Ellis & Young, 1996).11 Although it is assumed that the phonological input and phonological output buffers are a necessary component of the

language-processing model, the current letter-processing model does not explicitly illustrate them.



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

representation and phoneme production of letterswas intact: He was able to repeat letter namesand sounds and nonwords; he was able to recitethe alphabet from memory; and he was able to dis-tinguish between recognizable letter names andother similar-length nonwords (that were notletter names).

However, ET was unable to perform letter/number categorization for auditory stimuli. Inorder to determine whether an auditory stimulusis a letter or a number one must access semantics(given that both letters and numbers have phono-logical representations). Thus, ET’s impairment inauditory letter/number categorization is consist-ent with semantic impairment.

2. Letter semantics

We take the view here that single letters havesemantic representations. We propose that seman-tic representation of letters includes a type of

modality-free encyclopaedic knowledge aboutletters and their characteristics, such as knowledgeof the alphabet, knowledge of letter subtypes (suchas vowels and consonants), and what letters are andwhat they are used for. Letter semantics are rarelydiscussed in the literature; however, not all authorsagree with us. For example, Saffran and Coslett(1998) argue that “letters . . . have no conceptualcomponent [nor] semantic content”. Moreresearch is needed to determine the validity ofour assumption.

Nevertheless, from the current data we wouldargue that ET’s semantic knowledge of letterswas severely impoverished. He was unable tocategorize letters as vowels or consonants and dis-played very little general knowledge about letters.In addition, as mentioned above, he was unable tomake an auditory distinction between letter andnumber names (in the context of otherwiseintact phonological representations of singleletters).

Table 5. Other pre- and posttreatment results: Proportion correct

Task Pretreatment Posttreatment N

Significance value

p (McNemar)

Spoken-letter sound to written-letter matching

(PALPA 23)

0.50 0.88 26 .013

Writing letters to dictation of letter sounds 0.23 0.81 26 .000

Cross-case copying 0.0 0.50 20 .002

Cross-case copying of letters produced in free

writing

0.25a 0.69b — .063c

High-frequency regular word reading 0.0 0.93 15 .000

Spoken-to written-word matching task

(target words correct)

0.4 0.90 10 .063

Free writing (raw score, number of real words) 4 9 — NA

Letter naming (PALPA 22)

Upper case

B1 0.27 .35d 26 .688

B2 0.12

Lower case

B1 0.23 .42d 26 .07

B2 0.15

General knowledge for letters

Forced-choice questions 0.46 0.54 13 1.0

aN¼ 12. bN¼ 16. cAnalysis employed N¼ 10 matched items only. dET’s usual initial response was the letter sound (as he had been

trained only on letter sounds not letter names). This score includes those items for which ET responded correctly after prompting

for the letter name.



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

3. Recognizing written letters (and words)

Visual analysis systemThe visual analysis stage of processing is commonto visual object, face, and word processing domains(Riddoch & Humphreys, 1993). For a letterstimulus, visual analysis involves perception ofthe shape, position, and orientation of componentletter features and the integration of these featuresto form a complete letter percept (Balota, 1994;Coltheart, 1987; Massaro, 1998; Massaro &Sanocki, 1993). An individual with intact visualanalysis will be able to form a complete perceptof an individual letter that is font and case specificand thus copy, match, and discriminate betweenidentical letters of the same case and font.

Thus, in reference to Figure 6, ET’s visualanalysis system was relatively intact. He was ableto match and copy identical letter stimuli withease and also represent letter orientation relativelywell.

Abstract visual letter representationIt is common in models of word processing tocombine abstract visual representations of letterfont and letter case into a single module of abstractletter representation (e.g., Bigsby, 1988, 1990;Coltheart, 1981; Jackson & Coltheart, 2001; Kayet al., 1992; Thompson, 1999). The frameworkshown in Figure 6 explicitly separates these intotwo levels, labelled font-free representation andcase-free representation.

Font-free representation. At this level of processingEnglish letters of all fonts are represented as asingle case-specific abstract letter form (e.g., gand g share a single abstract representation forlower case g, while G and G share a separate rep-resentation for upper case G). Note that this isdifferent from allographic letter forms (which arefont-specific letter forms, i.e., the variant shapesof a letter). Individuals with intact font-free rep-resentations will be able to match two same-caseletters, even when presented in contrasting fonts(e.g., g and g, or a and a), or contrasting handwrit-ing styles. They will also be able to perform nor-mally on a visual letter/pseudoletter decision task

(i.e., distinguish between visually presentedletters and nonalphanumeric distractors). This isexactly the pattern shown by ET, and hence weargue that ET’s representations at the font-freelevel were relatively intact.

Case-free representation. At this level upper andlower case letters (e.g., the letters G and g) are rep-resented by a single abstract representation. Case-free representations are abstract representations ofsingle letters. Note that our use of the term “letter”is deliberate. This is not to be confused with theterm “grapheme”, which we define as the writtenrepresentation of a phoneme (Coltheart, 1978).For example, the letter string BREACH has sixletters but only three graphemes. We proposethat abstract case-free representations are ofsingle letters, not graphemes.

There is considerable research evidence sup-porting this level of visual representation (e.g.,Arguin & Bub, 1995; Bigsby, 1988; Coltheart,1981). An individual with intact case-free rep-resentation will be able to complete a cross-casematching task. ET’s cross-case matching wasseverely impaired, indicating poor developmentof case-free representations (or access to them).

Abstract visual letter representation and letterdecision tasksWe propose that font-free representations can bedisrupted independently of case-free represen-tations (as demonstrated by ET’s assessmentresults) and that font-free representations cansupport successful performance on letter decisiontasks involving nonalphanumeric distractors.

Visual letter decision with nonalphanumericdistractors. Arguin and Bub (1995) use a primingparadigm to demonstrate that, in normal proces-sing, alphabetic decision (identification of a letteramong nonalphanumeric distractors such as # ?%) can occur via access to a representation that isspecific to letter shape. Their results suggest thatthere is a level of visual letter representation thatis case specific and that these representations aresufficient for successful alphabetic decision per-formance. We would propose that activation of



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

Figure 6. Letter-processing framework.



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

case-specific visual letter representations wouldalso be sufficient to distinguish “real” Englishletters from other unfamiliar pseudoletters (e.g.,( s S).12 Our proposal is also supported byreported cases for whom cross-case matching isimpaired but alphabetic decision remains intact(Greenwald & Gonzalez Rothi, 1998; Miozzo &Caramazza, 1998).

Visual letter decision with numeric distractors. EThad intact font-free representations and thus wasable to distinguish between visually presentedEnglish letters and pseudoletters. However, hecould not distinguish between visually presentedletters and numbers. We therefore propose thatvisual number forms are represented at the font-free (case-specific) level of processing, and thusletter/number decision cannot be performed suc-cessfully at this level. In addition, given that onlyletters are represented at the case-free level of pro-cessing (numbers do not vary in case) we wouldargue that the case-free level of representationcan distinguish between visually presented lettersand numbers. Thus, ET’s poor performance onvisual letter/number decision (categorization) isconsistent with a core impairment of case-freerepresentation.

Orthographic lexiconThe orthographic lexicon represents visual rep-resentations of letter strings that we have learnedto recognize (i.e., familiar words). The mostwidely accepted test of integrity of the ortho-graphic lexicon is the visual lexical decision task,which requires visual discrimination betweenwritten words and nonwords. If single letterswere included as stimuli in the administration ofthis task they would be rejected (e.g., is B aword?). We would therefore argue that the ortho-graphic lexicon operates at the word level and con-tains no representations of single letters (except forthose that are words, e.g., A, I).

ET’s performance on a high-frequency wordlexical decision task was very poor. We wouldtherefore argue that his development of ortho-graphic representations was severely delayed.

Case-free bypass routeCurrently most theories of reading assume thatnormal skilled reading is performed in a hierarch-ical manner. That is, word recognition depends onsuccessful identification of component letters. Ithas also been assumed that “the computation ofabstract letter identities is a precursor to lexicalaccess during reading” (Besner et al., 1984,p. 126). One prediction from this theoretical pos-ition would be that individuals with no access toabstract letter representations (case-free represen-tations) will be unable to recognize or read aloudany words. However, cases have been reportedfor whom word reading remains possible despitevery impaired access to case-free representationof letters (e.g., Howard, 1987; Lambon Ralph &Ellis, 1997; Rapp, Link, & Caramazza, 1993,cited in Rapp, Folk, & Tainturier, 2001). Hence,we propose that font-free (and case-specific)representations of letters can be sufficient for acti-vation of word representations in the orthographicinput lexicon (as shown by the case-free bypasspathway, Pathway 1, Figure 6).

Impaired access to case-free representations andresidual reading skills. Howard (1987) reports onTM, a case of acquired deep dyslexia with severelyimpaired letter processing. TM’s cross-casematching was consistently at chance; however,TM was able to match letters presented in differ-ent fonts relatively well. According to Figure 6,one would predict that because TM’s font-freerepresentations were relatively intact he would beable to access some orthographic representations,despite his very impaired ability to access case-free representations. This was in fact true forTM. His word reading remained partially possible(i.e., 30–40% of words were read correctly).

12 We are not proposing that letter decision tasks can only be completed successfully at the level of font-free representation, just

that this level of representation would be sufficient.



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

Therefore, case TM provides evidence in supportof the case-free bypass route.13

Impaired access to case-free representations: Why isn’treading perfect?. If we are proposing that the case-free bypass route allows direct access to ortho-graphic information, we also need to considerwhy word recognition is not perfect in cases withimpaired performance on a cross-case matchingtask. We are not aware of any reported case withimpaired access to case-free representations (butintact font-free representations) that performsnormally on visual lexical decision (for words)and/or reading aloud. For example, visual lexicaldecision accuracy was impaired for FRE (Wilson,White, & McGill, 1983, cited in Wilson,1987), GV (Miozzo & Caramazza, 1998), MR(Greenwald & Gonzalez Rothi, 1998), and TM(Howard, 1987).

To explain the impaired lexical decisionperformance of these cases we would have toargue for an additional impairment either affectingthe orthographic lexicon itself or access to it via thebypass route. However, for GV and MR ortho-graphic representations were intact as shown byintact recognition of orally spelled words.14

Hence, we could argue that these cases have anadditional impairment to the bypass pathway,resulting in poor lexical decision.

An alternative account is that the case-freebypass route relies on an inefficient global formof word processing. Howard (1987) argues thatthe bypass route is an inefficient reading route inwhich words are processed globally as picturesrather than analysed according to the sequence ofindividual letters. He argues that the “globalword recognition route” is most useful in situationswhere top-down processing provides abundantinformation and context, so that even if the

reader fails to analytically identify a word theirexpectations about it can be quickly confirmed.Howard (1987) proposes that skilled readersemploy both the traditional lexical reading route(via identification of abstract letter forms) andthe “global word recognition route” for readingdepending on the demands of the reading task.However, cases with very impaired case-freeletter representations (such as TM) are forced touse it for single-word reading, where its usewould normally be inappropriate.

The case-free bypass route and ET. ET was unable toread words aloud and performed poorly on visualword lexical decision (employing high-frequencywords). With reference to Figure 6, ET hadnever been able to access case-free representationsand was thus forced to solely employ the case-freebypass route to acquire reading skills. The ineffi-ciency of this route, in conjunction with semanticdifficulties (i.e., limited top-down processing)seems adequate to explain his failure to developorthographic representations.

4. Writing letters

Orthographic bufferIt is widely accepted that an abstract represen-tation of letter identity exists at the level of theorthographic buffer (e.g., Caramazza & Miceli,1990; Ellis, 1982; Goodman & Caramazza,1986). This representation is case free and fontfree and multimodal (common to both writtenand oral spelling).

Peripheral letter shape conversionThere is some debate among authors regardingthe exact number and/or nature of peripheralrepresentations in the writing process (Del

13 Miozzo and Caramazza (1998) suggest that TM’s residual reading ability results from a guessing strategy supported by limited

access to orthographic information. If so, one would also expect a length effect in reading aloud—that is, better reading accuracy with

longer words than shorter words due to greater orthographic distinctiveness of longer words (fewer orthographic neighbours).

However, Howard’s (1987) analyses indicate no significant relationship between word length and TM’s reading accuracy or latency.14 Words were presented to the auditory modality in the form of letter names. Recognition was demonstrated via intact word

naming or lexical decision performance.



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

Grosso Destreri et al., 2000; Rapp & Caramazza,1997; Tainturier & Rapp, 2001).15 However,there is general agreement that the peripheralwriting system enables us to convert abstractorthographic representations into more specificletter forms. A popular framework (e.g., Ellis,1982, 1988; Goodman & Caramazza, 1986)posits two peripheral levels of processing specificto handwriting: the allograph level and the gra-phomotor level. The allograph level is conceivedof as a store of visuospatial representations ofletters according to their shape (i.e., a visual rep-resentation of letter shape that is specific to caseand style, S! S, s, ). The graphomotor levelallows selection of graphic motor patterns (basedon the pattern of allographic activation) thatspecify the sequence of written strokes requiredto write the selected letter.

In essence, the current framework divides theallograph level of representation into two separatelevels corresponding to case selection and fontselection (see Figure 6), based on evidence of adouble dissociation between and within theselevels of representation (Hanley & Peters, 1996;Venneri, Pastell, & Caffarra, 2002; Weekes,1994). Graphomotor representations and moreperipheral motor processes are not shown.

Case selection. Impairment in case selection is pro-posed to result in case errors in writing. Individualswith acquired dysgraphia have been reported whoproduce mixed-case errors when writing (e.g.,De Bastiani & Barry, 1989), or who exhibit sig-nificantly more writing impairment in lower thanin upper case (e.g., Graham, Patterson, &Hodges, 1997; Hanley & Peters, 1996;Kartsounis, 1992; Patterson & Wing, 1989;Weekes, 1994) or vice versa (e.g., Del GrossoDestreri et al., 2000; Trojano & Chiacchio, 1994).

Font selection. Selective impairments in font selec-tion have also been reported in the literature.Cases have been reported with preserved ability

to write in print but an inability to producewriting in cursive style (e.g., Venneri et al.,2002) and vice versa (Hanley & Peters, 1996;Venneri et al., 2002).

In reference to written-letter production, ETwas able to complete partial letter forms, wasable to produce a number of letters in freewriting plus more in other writing to dictationtasks, and could copy English letters faster thanGreek letters. We would therefore argue (withreference to Figure 6) that his retrieval of letterrepresentations from the orthographic buffer andproduction of specific written-letter shapes wererelatively intact.

Cross-case copyingCross-case copying has been represented inFigure 6 by the direct route from case-free rep-resentation to the orthographic buffer (pathway2). We propose that cross-case copying relieson intact case-free visual representation ofletters, and thus impairments in cross-case match-ing will result in impaired cross-case copying (e.g.,case GV, Miozzo & Caramazza, 1998). Thisproposal could be disputed by report of an individ-ual with impaired cross-case matching but intactcross-case copying. We are not aware of such acase.

Given ET’s relatively good written-letterproduction (as just discussed), his failure oncross-case copying tasks most likely reflects hissevere impairment at the level of case-free visualrepresentation (which prevents successful proces-sing via pathway 2).

5. Letter naming

Given the assumption that single letters are notrepresented in the orthographic lexicon, wepropose that letter naming can be achieved via adirect link between visual representations ofletters and the phonological lexicon (see pathway3, Figure 6). Without pathway 3 the system

15 Readers are referred to Tainturier and Rapp (2000) for a concise summary of central and peripheral spelling processes,

including current debates.



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

would not be able to name visually presentedletters. Based on evidence from priming studies(Arguin & Bub, 1995) we argue that letternaming requires the activation of case-free rep-resentations. Thus we argue that ET’s letter-naming impairment resulted from his impairedcase-free representations, given intact phonologi-cal representations of letters (including intactspoken-letter production).

Abstract letter representations and letter namingBased on the proposal that letter naming relies onactivation of case-free visual letter representations,an individual with very impaired cross-case match-ing should always have impaired letter naming. Toour knowledge, most of those reported cases withvery impaired cross-case matching also have veryimpaired letter naming (e.g., Greenwald &Gonzalez Rothi, 1998; Howard, 1987; Miozzo& Caramazza, 1998; Wilson, White, & McGill,1983, cited in Wilson, 1987). More importantly,visual letter naming is impaired even in thosecases with intact letter name production (asdemonstrated through intact oral spelling, whichrequires the production of spoken letter names;Greenwald & Gonzalez Rothi, 1998; Miozzo &Caramazza, 1998). For reported cases whosecross-case matching is poor but letter-namingability is relatively good, results are confoundedby difficulties with early visual analysis and variableperformance on cross-case identification tasks(AB, Lambon Ralph & Ellis, 1997) or above-chance performance on cross-case matching tasks(JGE, Rapp et al., 1993, cited in Rapp et al.,2001).

Finally, if letter naming does not rely on case-free representations and could be supported byfont-free visual representations we might expectsome individuals with acquired dyslexia to havecase-specific naming difficulties. To our knowl-edge no such case has been reported.

Thus we would argue that research evidence atthis time provides support for our proposal that

access to case-free representations of letters isnecessary for successful letter naming.

6. Letter sounding

Letter sounding can occur through the use ofgrapheme to phoneme(s) conversion rules (seeFigure 6, pathway from case-free representationto phoneme production). ET’s letter soundingwas severely impaired, presumably due to impairedcase-free representations (given intact phonologi-cal processing of single letters).

Abstract letter representations and nonword readingWe propose that nonword reading relies on acti-vation of case-free visual letter representations.This proposal could be disputed by report of anindividual with very impaired cross-case matchingbut intact letter sounding or nonword reading.None of MR (Greenwald & Gonzalez Rothi,1998), TM (Howard, 1987), GV (Miozzo &Caramazza, 1998),16 or AB (Lambon Ralph &Ellis, 1997), all cases with impaired access tocase-free representation of letters, could read non-words aloud. More importantly, even for thosecases with intact phonological output systems(intact oral spelling using spoken letter names),nonword reading was impaired (Greenwald &Gonzalez Rothi, 1998; Miozzo & Caramazza,1998). Thus the overwhelming evidence at thisstage supports the proposal that sublexical proces-sing relies of the formation of abstract case-freeletter representations.

7. Writing letters to dictation of namesand sounds

Writing letters to dictation from sounds can beachieved using phoneme to grapheme conversion(see Figure 6, via the auditory analysis system,phoneme production system, phoneme to gra-pheme conversion, orthographic buffer, and caseand font selection modules). We propose thatwriting letters to dictation of names occurs via a

16 Although nonword reading aloud was not specifically tested, word reading (including regular words) was impossible.



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

direct link between the phonological lexicon andorthographic buffer (see pathway 4, Figure 6).Pathway 5 represents the process of oral spelling,which is not relevant to the current case.

ET displayed a marked impairment in writingletters to dictation of names and sounds. GivenET’s ability to produce in free writing letters thathe cannot produce to dictation, this is unlikely tobe the result of an orthographic buffer impairment.So, given relatively intact functioning at the levelof the phonological lexicon, phoneme productionsystem, and orthographic buffer (for singleletters), the most likely causal impairment is ageneral failure to develop or learn the associationsbetween letter names/sounds and their graphemicforms, implicating both grapheme to phonemeconversion and pathway 4 (refer to Figure 6).

8. Evaluating the letter-processing framework

The current framework proved successful inexplaining ET’s pattern of letter-processingimpairment. In brief, ET demonstrated threecore impairments affecting case-free represen-tation of letters, letter semantics, and connectionsbetween the phonological processing modules andorthographic buffer. In our discussions about theframework we have made some explicit proposalsabout abstract letter identities (or case-free rep-resentations) and their relation to letter proces-sing, word processing, and nonword processingin this paper. We have proposed that:

1. Activation of case-free visual representations ofletters is necessary for successful performanceon cross-case copying, letter naming, lettersounding, and nonword reading.

2. Activation of font-free (and case-specific) rep-resentations of letters is sufficient for activationof word representations in the orthographiclexicon (as shown by pathway 1, Figure 6).

We hope that these proposals will be evaluatedfurther in future research. Cases with only mildimpairment in access to abstract letter identities(e.g., Patterson & Kay, 1982; Perri, Bartolomeo,& Silveri, 1996) or inefficiency in processing(e.g., Arguin & Bub, 1994; Behrmann &

Shallice, 1995; Kay & Hanley, 1991;Reuter-Lorenz & Brunn, 1990) are not ideal fortesting the above proposals. The most ideal casesare those for whom cross-case matching perform-ance is consistently no better than chance.Unfortunately this level of impairment is rarelyreported (Howard, 1987; Miozzo & Caramazza,1998).

9. Development of abstract letterrepresentations

There are very few investigations that specificallyinvestigate the development of abstract letter iden-tities (i.e., case-free letter representations) innormal children or developmental dyslexia(Bigsby, 1990; McFarland, Frey, & Landreth,1978; Rynard & Besner, 1987). Bigsby (1990) inher review of the literature found no cognitiveexperimental analysis of developmental dyslexiathat included specific assessment of abstract letterrepresentation. Bigsby’s (1990) findings (basedon primed reaction time tasks) suggest a normaldevelopmental trend towards greater processingefficiency or automaticity of access to case-freevisual letter representations (also see McFarlandet al., 1978). In addition, her analysis of a groupof developmental dyslexics revealed that abstractletter coding deficits are not uncommon in poorreaders. We would endorse Bigsby’s (1990) rec-ommendation for more systematic investigationof abstract letter representation in developmentaldyslexics given that delayed development ofabstract letter codes may represent a core impair-ment for some developmental dyslexics.

In adults, there is evidence that abstract lettercodes can be accessed without reference to a pho-nological code (e.g., Besner et al., 1984; Bigsby,1988; Coltheart, 1981). But, can childrendevelop abstract letter representations withoutbeing explicitly taught their names? Rynard andBesner (1987) investigate this proposal in a casestudy of PH, a 16-year-old developmental dys-lexic. His reading was at a Grade 1 level(i.e., very minimal reading skills), and letternaming and sounding were impaired (69% and65% correct respectively). In contrast, cross-case



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

matching was perfect, even for those letters that hewas unable to name/sound. In addition, in a timedcross-case matching task, letters that PH correctlynamed were matched no more quickly than lettersthat he was unable to name. Thus, the relationshipbetween upper and lower case letter forms can bebased on a purely visual link between symbols.The acquisition of cross-case matching skillsdoes not depend on children learning the equival-ence of two symbols (e.g., A a) through assign-ment of a common name or sound.17

10. Treatment generalization tountreated items

Treatment methods for ET focused on case-freeand semantic representation of letters and lettersounding. Treatment resulted in dramatic andenduring improvements in all three outcomemeasures: letter sounding, cross-case matching,and letter/number categorization. In addition,treatment generalization to untreated items wasalso apparent, as discussed below.

Cross-case matchingFor cross-case matching, assessment results (at mid-treatment assessment) indicate generalization oftreatment effects to untreated items. This is anunexpected but interesting finding. Until we knowmore about how abstract letter representations areformed during development it is difficult to beclear about how this might have occurred. PerhapsET’s case-free representations were partiallyformed prior to treatment, and treatment impactedon access to all of them in some nonspecific manner.

Letter/number categorizationGeneralization to untreated items was also evidentin letter/number categorization. ET’s improve-ment in visual categorization of letters versusnumbers is consistent with his better accessto case-free representations at midtreatment.Improvement in the auditory categorization task

suggests improved access to semantic informationabout categorization of numbers and letters. Thisprovides some additional evidence that the currentintervention, which was designed to target semanticrepresentations (in addition to case-free represen-tations), did result in some semantic changes.

Letter soundingWith reference to Figure 6, we might predict thatimproved access to case-free letter representationswould in turn allow more successful processing viathe grapheme to phoneme conversion route.However, ET’s results indicate no generalizationof letter sounding to untreated letters despite theimprovement in case-free representations forthese letters.

ET’s letter-processing difficulties were develop-mental. He had never had normal access to case-free representations prior to treatment and wastherefore unable to acquire grapheme tophoneme conversion rules. Although at midtreat-ment access to case-free representations hadimproved we would not predict automaticimprovement in letter sounding without explicitinstruction in grapheme to phoneme associations.

FINAL COMMENTS

First, in terms of clinical implications, the currentpaper describes a treatment method for remedia-tion of severe developmental letter-processingimpairments within a cognitive neuropsychologi-cal framework. The treatment methods weresuccessful, and resultant improvements in letterprocessing were enduring.

Second, in terms of research implications, thiscase study has highlighted the relative lack oftheoretically based research investigating single-letter processing in a multimodal manner. Thepreliminary framework proposed in this paperattempts to integrate single-letter processes intocurrent models of language processing. The

17 A colleague (Sara Coombes) has provided us with further supporting data, from assessment of a 7-year-old child with dyslexia.

The child found letter naming and sounding (and matching written letters to sounds) and reading aloud impossible but cross-case

matching was perfect (in the context of normal spoken-language skills).



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

framework is employed successfully to interpretdata arising from the current case and other pub-lished case reports. We trust that our attempt todevise a letter-processing framework with explicittestable predictions will provide a platform forfurther research. Finally, we hope that our findings(in conjunction with those of Bigsby, 1990) willencourage researchers to investigate the integrityof case-free representations (i.e., cross-case match-ing abilities) as a possible underlying impairmentin severe developmental dyslexia.

Manuscript received 14 December 2004

Revised manuscript received 11 August 2005

Revised manuscript accepted 17 August 2005

First published online 15 February 2006

REFERENCES

Arguin, M., & Bub, D. (1994). Pure alexia: Attemptedrehabilitation and its implications for interpretationof the deficit. Brain and Language, 47, 233–268.

Arguin, M., & Bub, D. (1995). Priming and responseselection processes in letter classification andidentification tasks. Journal of Experimental

Psychology: Human Perception and Performance, 21,1199–1219.

Balota, D. (1994). Visual word recognition: The journeyfrom features to meaning. In M. Gernsbacher (Ed.),Handbook of psycholinguistics. San Diego, CA:Academic Press.

Behrmann, M., & Shallice, T. (1995). Pure alexia:A nonspatial visual disorder affecting letteractivation. Cognitive Neuropsychology, 12, 409–454.

Besner, D., Coltheart, M., & Davelaar, E. (1984). Basicprocesses in reading: Computation of abstract letteridentities. Canadian Journal of Psychology, 38,126–134.

Bigsby, P. (1988). The visual processor module andnormal adult readers. British Journal of Psychology,79, 455–469.

Bigsby, P. (1990). Abstract letter identities and develop-mental dyslexia. British Journal of Psychology, 81,227–263.

Bishop, D. V. M. (1983). The test for reception of

grammar. (Available from Age and CognitivePerformance Research Centre, University ofManchester, M13 9PL.)

Boles, D., & Clifford, J. (1989). An upper- and lower-case alphabetic similarity matrix, with derived gener-ation similarity values. Behaviour Research Methods,

Instruments, & Computers, 21, 579–586.Brunsdon, R., Nickels, L., Coltheart, M., &

Joy, P. (2004) Assessment and treatment of childhood

topographical disorientation: A case study. Manuscriptsubmitted for publication.

Caramazza, A., & Miceli, G. (1990). The structure ofgraphemic representations. Cognition, 37, 243–297.

Cole-Virtue, J. C., & Nickels, L. A. (2004). Whycabbage and not carrot?: An investigation of factorsaffecting performance on spoken word to picturematching. Aphasiology, 18, 153–180.

Coltheart, M. (1978). Lexical access in simple readingtasks. In G. Underwood (Ed.), Strategies of infor-

mation processing. London: Academic Press.Coltheart, M. (1981). Disorders of reading and their

implications for models of normal reading. Visible

Language, 15, 245–286.Coltheart, M. (1987). Functional architecture of the

language-processing system. In M. Coltheart,G. Sartori, & R. Job (Eds.), The cognitive neuropsy-

chology of language. Hove, UK: Lawrence ErlbaumAssociates Ltd.

Coltheart, M., & Leahy, J. (1996). Assessment of lexicaland nonlexical reading abilities in children: Somenormative data. Australian Journal of Psychology, 48,136–140.

Coltheart, M., Rastle, K., Perry, C., Langdon, R., &Ziegler, J. (2001). DRC: A dual route cascadedmodel of visual word recognition and readingaloud. Psychological Review, 108, 204–256.

De Bastiani, P., & Barry, C. (1989). A cognitive analysis ofan acquired dysgraphic patient with an “allographic”writing disorder. Cognitive Neuropsychology, 6, 25–41.

Del Grosso Destreri, N., Farina, E., Alberoni, M., Pomati,S., Nichelli, P., & Mariani, C. (2000). Selective uppercase dysgraphia with loss of visual imagery of letterforms: A window on the organisation of graphomotorpatterns. Brain and Language, 71, 353–372.

D’Ydewalle, G., & Auwers, T. (1994). Orthographicredundancy in letter recognition: Orthographicneighbourhood or orthographic context. European

Journal of Cognitive Psychology, 6, 287–310.Elliott, C. (1990). Differential Ability Scales manual. San

Antonio, TX: The Psychological Corporation.Ellis, A. W. (1982). Spelling and writing (and reading

and speaking). In A. Ellis (Ed.), Normality and

pathology in cognitive functions. London: AcademicPress, Inc.



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

Ellis, A. W. (1988). Modelling the writing process. InG. Denes, C. Semenza, P. Bisiacchi, &E. Andreewsky (Eds.), Perspectives in cognitive

neuropsychology. Hove, UK: Lawrence ErlbaumAssociates Ltd.

Ellis, A., & Young, A. (1996). Human cognitive neurop-

sychology: A text book with readings. Hove, UK:Psychology Press Ltd.

Gathercole, S., & Baddeley, A. (1996). The children’s test

of nonword repetition. London: The PsychologicalCorporation Ltd.

Golden, R. (1986). A developmental neural model ofvisual word perception. Cognitive Science, 10, 241–276.

Goodman, R., & Caramazza, A. (1986). Dissociation ofspelling errors in written and oral spelling: The roleof allographic conversion in writing. Cognitive

Neuropsychology, 3, 179–206.Graham, N., Patterson, K., & Hodges, J. (1997).

Progressive dysgraphia: Co-occurrence of centraland peripheral impairments. Cognitive Neuro-

psychology, 14, 975–1005.Greenwald, M., & Gonzalez Rothi, L. (1998). Lexical

access via letter naming in a profoundly alexic andanomic patient: A treatment study. Journal of the

International Neuropsychological Society, 4, 595–607.Hanley, J., & Peters, S. (1996). A dissociation between

the ability to print and write cursively in lower-caseletters, Cortex, 32, 737–746.

Howard, D. (1987). Reading without letters? InM. Coltheart, G. Sartori, & R. Job (Eds.), The

cognitive neuropsychology of language. Hove, UK:Lawrence Erlbaum Associates Ltd.

Jacewicz, M. (1979). Word context effects on letterrecognition. Perceptual and Motor Skills, 48, 935–942.

Jackson, N., & Coltheart, M. (2001). Routes to reading

success and failure. Hove, UK: Psychology Press.Jacobs, A., & Grainger, J. (1991). Automatic letter

priming in an alphabetic decision task. Perception

and Psychophysics, 49, 43–52.Kartsounis, L. (1992). Selective lower-case letter

ideational dysgraphia. Cortex, 28, 145–150.Kay, J., & Hanley, R. (1991). Simultaneous form

perception and serial letter recognition in a case ofletter-by-letter reading. Cognitive Neuropsychology,8, 249–273.

Kay, J., Lesser, R., & Coltheart, M. (1992). Psycholinguistic

Assessments of Language Processing in Aphasia (PALPA).Hove, UK: Lawrence Erlbaum Associates Ltd.

Korkman, M., Kirk, U., & Kemp, S. (1997). NEPSY. A

developmental neuropsychological assessment. SanAntonio, TX: The Psychological Corporation.

Lambon Ralph, M., & Ellis, A. (1997). “Patterns ofparalexia” revisited: Report of a case of visualdyslexia. Cognitive Neuropsychology, 14, 953–974.

Manly, T., Robertson, I., Anderson, V., &Nimmo-Smith, I. (1999). The test of everyday

attention for children (TEA-Ch). Bury St. Edmunds,UK: Thames Valley Test Company.

Manson, J., & Wendon, L. (1997). Letterland early years

handbook. Cambridge, UK: Letter LandInternational.

Massaro, D. (1998). Models for reading letters andwords. In D. Scarborough & S. Sternberg (Eds.),Methods, models, and conceptual issues: An invitation

to cognitive science: Vol. 4. An invitation to cognitive

science. Cambridge, MA: MIT Press.Massaro, D., & Hary, J. (1986). Addressing issues in letter

recognition. Psychological Research, 48, 123–132.Massaro, D., & Sanocki, T. (1993). Visual information

processing in reading. In D. Willows, R. Kruk, &E. Corcos (Eds.), Visual processes in reading and

reading disabilities. Hove, UK: Lawrence ErlbaumAssociates Ltd.

McClelland, J., & Rumelhart, D. (1981). An interactiveactivation model of context effects in letterperception: Part I. An account of basic findings.Psychological Review, 88, 375–407.

McFarland, C., Frey, T., & Landreth, J. (1978). Theacquisition of abstract letter codes. Journal of

Experimental Child Psychology, 25, 447–458.Miozzo, M., & Caramazza, A. (1998). Varieties of pure

alexia: The case of failure to access graphemicrepresentations. Cognitive Neuropsychology, 15,203–238.

Morrison, I., & Butler, B. (1986). Set-intersectionmodel of letter recognition in the spatialprobe task. Canadian Journal of Psychology, 40,136–160.

Mozer, M. (1989). Types and tokens in visualletter perception. Journal of Experimental


Patterson, K., & Kay, J. (1982). Letter-by-letterreading: Psychological descriptions of a neurologicalsyndrome. Quarterly Journal of Experimental

Psychology, 34A, 411–441.Patterson, K., & Wing, A (1989). Processes in

handwriting: A case for case. Cognitive

Neuropsychology, 6, 1–23.Perri, R., Bartolomeo, P., & Silveri, M. (1996). Letter

dyslexia in a letter-by-letter reader. Brain and

Language, 53, 390–407.



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

Polk, T., & Farah, M. (1997). A simple commoncontexts explanation for the development of abstractletter identities. Neural Computation, 9, 1277–1289.

Rapp, B., & Caramazza, A. (1997). From graphemesto abstract letter shapes: Levels of representationsin written spelling. Journal of Experimental


Rapp, B., Folk, J., & Tainturier, M. (2001). Wordreading. In B. Rapp (Ed.), The handbook of cognitive

neuropsychology: What deficits reveal about the human

mind. Philadelphia, PA: Psychology Press.Reuter-Lorenz, P., & Brunn, J. (1990). A pre-lexical

basis for letter-by-letter reading. Cognitive

Neuropsychology, 7, 1–20.Riddoch, M., & Humphreys, G. (1993). BORB: The

Birmingham Object Recognition Battery. Hove, UK:Lawrence Erlbaum Associates Ltd.

Rynard, D., & Besner, D. (1987). Basic processes inreading: On the development of cross-case lettermatching without reference to phonology. Bulletin

of the Psychonomic Society, 25, 361–363.Saffran, E., & Coslett, H. (1998). Implicit vs. letter-by-

letter reading in pure alexia: A tale of two systems.Cognitive Neuropsychology, 15, 141–165.

Sanocki, T. (1988). Font regularity constraints onthe process of letter recognition. Journal of

Experimental Psychology: Human Perception and

Performance, 14, 472–480.Sanocki, T. (1991a). Intra- and interpattern relations in

letter recognition. Journal of Experimental Psychology:

Human Perception and Performance, 17, 924–941.

Sanocki, T. (1991b). Looking for a structural network:Effects of changing size and style on letterrecognition. Perception, 20, 529–541.

Sevush, S., & Heilman, K. (1984). A case of literalalexia: Evidence for a disconnection syndrome.Brain and Language, 22, 92–108.

Tainturier, M., & Rapp, B. (2001). The spelling process.In B. Rapp (Ed.), The handbook of cognitive neuropsy-

chology: What deficits reveal about the human mind.Philadelphia, PA: Psychology Press.

Thompson, B. (1999). The processes of learningto identify words. In G. Thompson &T. Nicholson (Eds.), Learning to read: Beyond

phonics and whole language. New York: TeachersCollege Press.

Trojano, L., & Chiacchio, L. (1994). Pure dysgraphiawith relative sparing of lower-case writing. Cortex,

30, 499–507.Venneri, A., Pestell, S., & Caffarra, P. (2002).

Independent representations for cursive andprint style: Evidence from dysgraphia inAlzheimer’s disease. Cognitive Neuropsychology, 19,387–400.

Wechsler, D. (1991). Manual for the Wechsler

Intelligence Scale for Children—Third Edition

(WISC-III). San Antonio, TX: The PsychologicalCorporation.

Weekes, B. (1994). A cognitive-neuropsychologicalanalysis of allograph errors from a patient withacquired dysgraphia. Aphasiology, 8, 409–425.

Wilson, B. (1987). Rehabilitation of memory. London:Guilford Press.



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

APPENDIX A

Errors in picture naming

Picture ET’s Response Error Type

glass cup semantic and visualrabbit kangaroo semantic and visualseal dolphin semantic and visualbelt collar semantic and visualcow bull semantic and visualswan duck semantic and visualonion watermelon semantic and visualmountain pinecone visualanchor bow and arrow visualiron phone visualarrow don’t know NAlemon don’t know NA

APPENDIX B

Daily home practice guide

Please complete the following with ET each day.

1. Read relevant page of the Letter Land ABC Book

2. Look at lower-case flash card - both sides (with picture cue

and without)

3. Play relevant song on CD. Sing song together.

4. Think of some words that start with the sound

e.g. “Can you think of some words that start with Harry

hat man’s sound?”

5. Introduce the capital letter

(i) Look at the capital letter flash card

(ii) Read the relevant capital letter script

(This will be provided for each letter)

6. Test ET’s knowledge of the characters and letter sounds

using the flash cards

(i) Start with the flash cards with picture

i.e., while showing the lower case flash card—

ask, “who is this”? and “what sound do they

make”? Then do the same with the upper case flash

card

(ii) Then do the other side (the letters without

pictures).

For the final day of each week: Revision of all the new letters

practised during the week.

. Look at the flash cards (both upper case and lower case

versions)

. Sing the song



Dow

nloa

ded

by [

McG

ill U

nive

rsity

Lib

rary

] at

08:

51 1

8 D

ecem

ber

2014

Documents

Severe developmental letter-processing impairment: A treatment case study